A&A 456, 783–788 (2006) Astronomy

DOI: 10.1051/0004-6361:20064941 &

c ESO 2006 Astrophysics

Synthetic representation of the Galilean satellites’ orbital motions

from L1 ephemerides

V. Lainey1,3 , L. Duriez2,3 , and A. Vienne2,3

1
Royal Observatory of Belgium, 3 avenue Circulaire, 1180 Bruxelles, Belgium
e-mail: Lainey@oma.be
2
Université Lille1 IMCCE/LAL, 1 impasse de l’Observatoire, 59000 Lille, France
3
Institut de mécanique céleste et de calcul des éphémérides–Observatoire de Paris, UMR 8028 du CNRS,
77 avenue Denfert-Rochereau, 75014 Paris, France

Received 31 January 2006 / Accepted 10 April 2006

ABSTRACT

Aims. We aim to provide a synthetic representation of the orbital motion of Io, Europa, Ganymede, and Callisto. It corresponds to
a quasi-periodic Fourier series whose arguments are integer combinations of the fundamental arguments of the dynamical system,
which are also given.
Methods. The present series are issued from frequency analysis and digital filtering treatments. The initial samplings have been taken
from a former numerical integration fit to the observations and referenced as L1 ephemerides.
Results. Our representation allows for the identification of all the significant perturbations in the system, which helps us to understand
these orbital motions. It is also useful in other related topics: the dynamical history of the system and studies requiring us to extract
only major terms to compute motions such as the rotation of the bodies.
Key words. ephemerides – celestrial mechanics – planets and satellites: general

1. Introduction
To have very accurate ephemerides of the Galilean satellites
Io, Europa, Ganymede, and Callisto, we built a new dynam-
ical model leading to numerical simulations of their motions
(Lainey et al. 2004a). This model has been adjusted to different
kinds of observations (photographic, CCD, and mutual events)
done between 1891 and 2003. Thousands of observations have
been used, including the most accurate ones, issued from the
mutual event campaigns (PHEMU) organized by the IMCCE,
from 1985 to 1992. As a result, we have given a numerical rep-
resentation that is able to produce ephemerides (Lainey et al.
2004b), referenced here as L1 ephemerides.
A numerical integration of the adjusted model was per-
formed over 1700 years. Instead of using a numerical repre-
sentation by Tchebychev polynomials, as is commonly made to
interpolate between time steps, we preferred using a frequency
analysis to build a quasi-periodic representation of the motions.
This method was first successfully used by Carpino et al. (1987)
in the case of planetary motions. Our representation has today a
faithful accuracy of a few tens or so kilometers upon one century
and remains definite upon more than 1700 years.
Furthermore, this synthetic form is fundamental for under-
standing the dynamics of the system. For example, the analyti-
cal representation of the eight main Saturnian satellites’ motion
(Vienne & Duriez 1995) has allowed us to discover new terms.
These terms lead to a better knowledge of the evolution of the
Mimas-Tethys resonance (Champenois & Vienne 1999a,b). For
the Galilean satellites, a representation rather similar to the one

In the present paper, we give the synthetic representa-
Tables 4 to 18 are only available in electronic form at
http://www.edpsciences.org
presented here has been used in the works of Henrard (2005a,b)
dealing with the rotation of Europa and Io. The present work has
also been used by Noyelles (2005) to describe the evolution of
the Galilean system. More generally, a synthetic form helps to
identify and understand the origin and effects of the perturba-
tions that occur in the motion of the corresponding satellites.
At the end of the 70s, Lieske (1977) refreshed Sampson
analytical ephemerides (Sampson 1921) for use in Voyager
spacecraft navigation. Despite the important works done by sev-
eral authors (e.g. Sagnier 1975; Vu 1982; Thuillot 1984), these
ephemerides are still probably the only ones that have both an
analytical model and a fit to the observations. However, the re-
lated representation suffer from a few inconsistencies: the num-
ber of parameters to fit was higher than the number of degrees
of freedom, and the Laplacian libration was artificially added in
the series. Recently, Lainey et al. (2004a) also pointed out that
some long period terms were missing in Sampson-Lieske the-
ory. Moreover, the representation was developed in cylindrical
variables instead of the usual elliptical elements. Hence, com-
parison between the present work and the Sampson-Lieske se-
ries is rather difficult. The present paper provides a representa-
tion of the Galilean system. We have avoided most difficulties
encountered by Sampson (1921) and Lieske (1977), thanks to
numerical integration. Because the numerical integration used
to simulate the satellite motions had previously been fit to the
observations, its present synthetic representation can be used for
ephemerides purposes. This synthetic formulation offers the ad-
vantage of describing the Galilean system over a very long time
span (as long as tidal effects and long solar period terms are con-
sidered negligible).
tion of the solution L1 and describe the method developed to

Article published by EDP Sciences and available at http://www.edpsciences.org/aa or http://dx.doi.org/10.1051/0004-6361:20064941

784 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions
obtain it. This representation corresponds to trigonometric se-
ries whose arguments are integer combinations of fundamental
arguments of the system: Li (the linear parts of mean longitudes
of the satellites and Sun), i and Ωi (the proper modes of the
pericenters and nodes), and ν, ρ, and Ψ (the arguments of great
inequalities and resonances). First, in Sect. 2, we study the fre-
quency spectrum of the Galilean dynamical system and charac-
terize their short and long periods. Section 3 presents how we
generated the different samplings we had to analyse. In Sect. 4
the frequency analysis algorithm we used is discussed. The next
section presents the filtering technique involved in separating the
long period from the short period. The last section is devoted to
the synthetic representation.
2. Form of the spectrum
A Fourier analysis usually requires a previous spectral knowl-
edge of the signal to determine optimal sample step size and time
span. We have used routines from Duriez (1990, 2003) adapted
to explore all the significant short period terms. For a given dy-
namical system of satellites (characterized by the satellite oblate-
ness and mutual perturbations), these routines determine the am-
plitude of a given inequality ki Li + k j L j , where Li and L j are the
linear parts of the mean longitudes of the satellites i and j, and
where ki and k j are integers. We recall that the subscript 1 is for
Io, 2 for Europa, 3 for Ganymede, and 4 for Callisto.
The computations are done at the first order with respect to
the masses. In these routines, the amplitudes are in fact a func-
tion of z = e exp i and ζ = sin 2I exp iΩ of each satellite (e, I,
, and Ω denoting, respectively, the eccentricity, the inclination,
the longitude of the pericenter, and the longitude of the node).
Because we only want an estimation of the amplitude of the in-
equality, we used the maximum values for the eccentricities and
the inclinations from Ferraz-Mello (1979). We explored all the
inequalities for which ki and k j were lower than 50 and for which
ki + k j was lower than 8 (the maximum degree in eccentricity-
inclination was then 8 because of the d’Alembert rule). We re-
tained all the short period terms with a truncation level of one
kilometer. One can find the list of those greater than 10 km in
Lainey (2002). It appeared that the short period term of signifi-
cant amplitude and longest period corresponds to the inequality
L3 − 2L4 (50, 2 days) and affects Ganymede. The significantly
shortest short period is higher than 0.6 day.
The previous routines may also be used for the long period
terms. Those involve the node and pericenter precessions (proper
elements), the great inequalities L1 − 2L2 = L2 − 2L3 + 180◦ de-
noted ν in the following text, the De Haerdtl inequality 3L3 −7L4
denoted ρ, the Laplacian libration argument Ψ, and the solar
Jovicentric motion1 . However, it was easier to explore these
terms by a preliminary frequency analysis of L1 (without the
filtering procedure described in Sect. 5.1), showing that no sig-
nificant long period term exists with a period below 160 days.
3. Sampling
To analyse L1 ephemerides, we used exactly the same software,
initial conditions and parameter values that were used in Lainey
et al. (2004b). The software computes and delivers Cartesian
1
As Saturn was also introduced indirectly in L1 modeling by means
of the planetary ephemerides DE406, its motion should also be con-
sidered here, in particular because of its great inequality in relation to
Jupiter. However, no identification introducing such frequency was pos-
sible, mainly because of the long period treatment (see Sect. 5.2).
positions and velocities of each satellite in a Jovicentric J2000
Earth mean equatorial frame. To obtain all samplings in the usual
elliptical elements, we first rotated each set of Cartesian coordi-
nates in a J2000 Jovian equatorial frame. Then, a routine was
applied to shift from Cartesian to Keplerian elements. This op-
eration was done for each data point.
4. Frequency analysis algorithm
The frequency analysis algorithm we used is very similar to
the one used in Vienne & Duriez (1992). We only indicate the
main points here; however, the reader can find more details in
the cited paper. First, a FFT is performed upon the initial sig-
nal determined by the numerical integration. We retain the fre-
quency corresponding to the highest amplitude from the FFT.
This frequency is only known with the precision ν0 = Tπ ,
where T is the time span of the numerical integration. ν0 is
called the fundamental frequency. Then, a fine analysis (Laskar
1993) is made in order to have the exact value of the frequency.
If this frequency is separated enough from the others, the pre-
√
cision reaches 1.24 ν0 (with the use of the Hanning win-
dow), where  is the epsilon-machine precision of the computer.
“Enough” means that the distance between two frequencies is
larger than 2ν0 . In that case, the corresponding term (frequency,
phase, and amplitude) is removed from the signal, and we ap-
ply the procedure again to find the next term. If, in one step of
the procedure, we find a frequency separated from a previously
determined one by less than 2ν0 , we put this term back in the
signal to determine it again. Sometimes it is enough, and the pro-
cedure continues with the discovery of new frequencies. But, if
the problem occurs again with the same term, then we determine
all the previous terms again, one by one. This method was firstly
used by Champenois (1999). He shows that, in some conditions,
the iterative procedure eliminates the problem of the influence
of one frequency on the others.
5. Synthetic analysis
In celestial mechanics, it is usually convenient to split the orbital
spectrum in two parts: the short period terms (typically a few
periods) and the long period terms (few pericenter and node pe-
riods). As discussed in Sect. 2, the Galilean system spectrum has
short periods that are less than a day and long periods of the or-
der of a few centuries. So, for the practical use of the frequency
analysis, it is better to cut the initial signal with respect to these
two parts. This method avoids manipulating a file of several hun-
dreds of megabytes: the time span T must be long enough to de-
termine all the long period terms, and the step size of the signal
must be small enough to determine the shortest period. We then
have to use digital filtering to obtain two signals for which the
sum of both is the original one. Indeed, the signal correspond-
ing to the short period terms can be set to a much smaller time
span than the original signal although the same step size can be
used. The signal corresponding to the long period terms can be
set with a much larger step size than the original signal, but with
the same time span.
5.1. Digital filtering
The digital filtering used here is very similar to the one used
in Duriez & Vienne (1997). The reader can find the details of
the procedure in that paper. In particular, we used a two-step
filtering process to avoid the use of a single filter with too high of
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions 785

Table 1. Characteristics of the two-step filtering used to remove the in all variables, except the semi-major axis. The recognition of
short period terms from the signal in the Galilean dynamical system. such terms would have needed a much longer time span, about
a few hundred thousand years, because of the period of revolu-
Filter 1 Filter 2 tion of the perihelies and nodes of the planets around the Sun.
∆t (days) 0.24 3.36 With a step size as short as a day, this would also increase the
T 0 (days) 6.72 60 numerical error in the integration process too much. Of course,
T 1 (days) 140 140 these very long period terms do not appear to be periodic over
δt (days) 3.36 26.88 1700 years only. To overcome this problem, in all variables ex-
2M + 1 231 247
cept the semi-major axis, we fit the residuals polluted by the
ρ 10−7 10−7
α 10−6 10−6 very long period terms with Tchebychev polynomials. An opti-
∆f 0.034 0.032 mal method was found in four steps. First, we filtered the residu-
als to remove undetected short period terms and kept only the
A very long period part. Then we fit the signal with low-order
Tchebychev polynomials (orders less than 8). Such polynomi-
als were then subtracted from the initial residuals, and we could
start our frequency analysis treatment again. In practice, few it-
ripple erations were enough to have a precision of a few tens of kilo-
1 ρ
transition band width meters. Figure 2 presents all four steps of this method applied to
of filter 1 the mean longitude of Callisto. To preserve space in this paper,
transition band
these Tchebychev polynomials are not given here, but are avail-
width of filter 2
able upon request to the authors. The long period terms found by
filter 2 frequency analysis in the mean longitudes (after subtracting the
filter 1
starting point starting point Tchebychev polynomials) have amplitudes greater than 2.2 km,
6.2 km, 6.1 km, and 21.8 km (respectively, in the mean longi-
tudes of Io, Europa, Ganymede, and Callisto). This truncation
level explains the precision of a few tens of kilometers that has
attenuation been obtained. In other variables, we were able to obtain trunca-
α tion levels half of those for mean longitudes.
2π /0.24 2π /3.36 2π/6.72 2π/60 2π /140 υ
Fig. 1. The characteristics of the two-step filtering.
5.3. Short period treatment
The short period spectrum is analysed by subtracting the recog-
nized long period terms on an unfiltered sample produced by
a number of coefficients. Table 1 summarises the characteristics our numerical integration. We can use a much less extended
of the two filters we used. Figure 1 visualises the characteristics time span since long periodic terms are removed. A step size
of the two-step filtering. ∆t denotes the step size in input for of 0.24 day was used and allowed us to detect all the short pe-
filtering, while δt is the output step size; [T 0 , T 1 ] and ∆ f are the riod terms longer than 0.48 day. A time span of 80 years was re-
cutting intervals expressed, respectively, in days and in sampling quired to separate the close frequencies of the system. We were
frequency; 2M + 1 is the filter length (odd number); and ρ and α able to adopt a truncation level of the short period terms that
denote the pass-band ripple and the stop-band attenuation. was less than that of long period ones (a level as low as about
This two-step filtering method was applied at the output of 0.2 km). Then, some residual long period terms were still able to
our numerical integration, after converting Cartesian coordinates be detected in the signal (at a level less than the amplitudes cited
into Keplerian elements. To filter the mean longitude signal, we above in the long period treatment). We again used the filtering
first subtracted a mean frequency empirically, and then refined it process with Tchebychev polynomials to progress in frequency
with a least squares fit. To obtain all long periodic terms, we sim- analysis up to this truncation level. However, the long period
ulated the Galilean satellites’ motion over +/–850 years, start- terms found in this way, along with the short period ones, seem
ing at the Julian epoch 2 433 282.5 (January 1st, 1950, at 0h). to be less reliable when trying to reconstruct their frequencies as
The numerical accuracy of such integration was estimated to combinations of the fundamental frequencies (see Sect. 6).
a few hundreds of meters. After filtering, our sample covered
1700 years with a step size of 26.88 days. This span is long
enough to detect the longest periodic term of the Galilean system 5.4. Precision
equal to 562 years (node of Callisto). The trigonometric series and Tchebychev polynomials ob-
tained by the above method represent what are called “L1
5.2. Long period treatment ephemerides”. The corresponding tables are available by ftp2 ,
as well as by a FORTRAN subroutine that can be used to com-
To analyse the long period terms of the Galilean system, we used pute ephemerides of the Galilean satellites. The subroutine out-
the frequency analysis method presented in Sect. 4 on the fil- put was compared to an unfiltered sampling delivered by our
tered signal. The efficiency of the quasi-periodic representation numerical model. The final accuracy (internal precision) over
was estimated by comparing the initial signal and the recon- one century (centred on initial epoch 2 433 282.5) of our fre-
structed one. We obtained good results for all elements (agree- quency representation was found to be of a few kilometers for
ment around few kilometers), except for the satellite mean lon- Io, 20 kilometers for Europa and Ganymede, and 35 kilometers
gitudes (a few tens of kilometers). Problems arise because of for Callisto. Over 1700 years (again centred on a J1950.0 epoch),
the presence of very long solar periodic terms that cannot be
2
analysed with only a 1700 year time span. These terms appear ftp://ftp.imcce.fr/pub/ephem/satel/galilean/L1
786 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions

Fig. 2. Four step method used to reconstruct Callisto’s mean longitude. X-axes are in years, and Y-axes are in kilometers. Figures on top present the
residuals before the Tchebychev treatment. On the left, a filtering process is performed to remove short period terms. On the right, a Tchebychev
polynomial is fit on the filtered signal. Figures on the bottom present the residuals after subtracting the Tchebychev polynomial (left) and applying
the frequency analysis treatment again (right).

our representation is accurate to 25 kilometers for Io, 90 kilome-
ters for Europa and Ganymede, and 150 kilometers for Callisto.
Such accuracy is just the internal precision of our representation.
The accuracy of L1 ephemerides (external precision) also intro-
duces the error coming from the fit process (mainly due to ob-
servational errors). Considering this, the global accuracy of L1
ephemerides can be estimated to 25 kilometers for Io, 40 kilo-
meters for Europa and Ganymede, and 55 kilometers for Callisto
over one century.
ρ: the De Haerdtl great inequality 3L3 − 7L4 .
LS : the linear part, in time, in the mean longitude of the Sun.
Except ν and ρ, which are connected to some Li , these arguments
represent the minimum set able to represent the full dynamics of
the Galilean satellites.
Table 2 gives the values of these arguments and the solutions
from which they issue. These solutions are given in Tables 3 to
18, where each term is in the form of a signed numerical co-
efficient that is supposed to be in the factor of a trigonometric
function (sine, cosine, or complex exponential) whose argument
corresponds to the formulae:

6. Identification and synthetic representation phase + frequency × T,

We first identified the fundamental arguments in the numeri-
cal representations issued from the frequency analysis process.
These arguments are:
Li : the linear part in the mean longitudes of the satellite i.
i : the four proper modes in the longitudes of the pericenters.
Ωi : the four proper modes in the longitudes of the nodes.
Ω0 : the proper mode related to the longitude of the node of the
Laplacian plane.
ν: the great inequality L1 − 2L2 or L2 − 2L3 + 180◦ .
We remind the reader that (L1 − 2L2 ) – (L2 − 2L3 ) = L1 −
3L2 + 2L3 oscillates around 180◦. This libration is the main ef-
fect of the Laplacian resonance.
Ψ: the argument of the Laplacian resonance.
where T = DJ − 2 433 282.5 (that is, the time from J1950, in
days). The time scale is TDB. The short period terms are pre-
sented together with the long period ones, simply by decreas-
ing amplitudes. These amplitudes are expressed in kilometers,
by a simple conversion of AU in km in the semi-major axis, or
by multiplying radians (in other variables) by the constant part
of the semi-major axis (in km) of the corresponding satellite.
The solutions are expressed in cosine for the semi-major axis ai ,
in sine for the mean longitudes Li , and in complex exponential
for the variables zi (eccentricity-pericenter) and ζi (inclination-
node).
The recognition of the fundamental arguments in these so-
lutions was possible because previous analytical works, such as
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions 787

Table 2. Fundamental arguments of the Galilean system used for the identification. Arguments are in the form Phase + Frequency × T with
T = JD − 2 433 282.5. The argument ν is also very close to the combinations L1 − 2 L2 and L2 − 2 L3 + 180◦ .

Fundamental Frequency Phase Period Period Issued from

argument (rad/day) (deg) (day) (year) solution
L1 3.551552286182 82.861918 1.769 0.005 L1
L2 1.769322711123 338.598517 3.551 0.010 L2
L3 0.878207923589 16.467319 7.154 0.019 L3
L4 0.376486233434 339.256972 16.689 0.046 L4
LS 0.001450183749 318.603037 4332.938 11.863 z4
Ψ 0.003050648672 109.319546 2059.622 5.639 L2
ν 0.012906864147 125.663720 486.809 1.333 z1 , z2 or z3
ρ –0.000779862000 194.603151 8056.791 22.058 = 3L3 − 7L4
1 0.002664253355 25.288760 2358.328 6.457 z1
2 0.000677973430 157.191935 9267.598 25.373 z2
3 0.000127274130 121.789605 49 367.340 135.160 z3
4 0.000032065099 319.917125 195 950.909 536.484 z4
Ω1 –0.002315096098 160.223177 2714.006 7.431 ζ1
Ω2 –0.000569206405 60.029150 11 038.500 30.222 ζ2
Ω3 –0.000124913071 191.189501 50 300.463 137.715 ζ3
Ω4 –0.000030561255 342.180346 205 593.170 562.883 ζ4
Ω0 0.000000000000 138.277188 ∞ ∞ ζ4

Table 3. Solution for the variable a1 (semi-major axis of Io). The series is expressed in cosine.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
422029.958 0.00000 0.0000000000
11.400 208.51597 3.5644591656 2L1 − 2L2 −0.155D-07 0.01083
2.706 57.04065 7.1289183312 4L1 − 4L2 −0.310D-07 0.01295
2.578 104.25820 1.7822295778 L1 − L2 −0.270D-08 0.00520
1.522 161.29083 8.9111478635 5L1 − 5L2 0.118D-07 0.02618
1.418 199.16142 8.0200331113 3L1 − 3L3 −0.235D-07 0.02238
1.379 265.54878 10.6933774362 4L1 − 4L3 0.142D-07 0.02962

that of Ferraz-Mello (1979) or of Duriez (1982) give, at least,
their period. It appears that the same fundamental argument is
present in several solutions, generally with very small differ-
ences (for instance 3 , in solutions z1 , z2 , z3 , or z4 ). We chose the
argument in one of these solutions that generally correspond to
the term whose amplitude is the largest, to have the best determi-
nation of the phase (for instance 3 was issued from solution z3 ).
After that, we tried to identify each frequency in the solu-
tions as integer combinations of the fundamental frequencies.
The redundancies of ν and ρ with the relating combinations of Li
where used to identify the frequencies in the long period part of
the solutions more easily.
In most cases, the frequency of each term is recognized as
an integer combination of the frequencies of the fundamental
arguments, by trying all possible combinations with increasing
integers until the best agreement is reached, but also by respect-
ing d’Alembert’s rule (except for the solar terms, see below).
As a consequence, higher integers for the arguments  and Ω
corresponds to higher degree in z and ζ (that is in eccentrici-
ties and inclinations). We then verify that the same combina-
tion of integer applied on the phases of the fundamental argu-
ments allow us to recover the phase of the term (a difference
of 180◦ was converted by changing the sign of the amplitude).
Tables 3 to 18 show how each term is identified as an inte-
ger combination of the fundamental arguments. In the two last
columns, we also give the difference in frequency and phase be-
tween their numerical values issued from the frequency analysis
process and their evaluation by the related combination of funda-
mental arguments. Identification is best when the differences are
smallest.
The solar terms are those depending on the argument LS , and,
to less of an extent, those with the argument Ω0 . Generally, these
terms also depend on the eccentricity and/or inclination of so-
lar Jovicentric motion, hence on the very long period associated
with the motions of the pericenter and/or node of the solar or-
bit. The time scale of the corresponding motion is much longer
than that of the satellite. Thus, the solar frequencies of the peri-
center and the node cannot be separated in our analysis. As a
consequence, we can recognize a frequency close to its mean
longitude frequency, but not the phase. This is why the values
in column “phase difference” are not significant for solar terms
and hence not given. Some other terms (essentially small terms)
are not well identified; these are considered as doubtful and in-
dicated by a star symbol (*) in the tables. These doubtful terms
come essentially from the long period part of the solution found
in the short period treatment (Sect. 5.3). The given identification
is considered doubtful when it corresponds to a very improb-
able inequality (like 2LS + ν − 3 − Ω3 in solution z2 or like
2LS − Ψ + 1 − Ω4 in the mean longitude of Io) or when the dif-
ference in frequency is larger than about 10−5 rad/day. For these
terms and for the solar terms, it is better to use the numerical
value of the frequency and the phase to evaluate the term instead
of its doubtful identification.
In Tables 3 to 18, we have used a truncation level of 1 kilo-
meter. There are some more terms with lower amplitudes, but
they are not really significant from an analytical point of view.
We recall that the internal precision of L1 is a few hundred of
meters. However, we give some smaller terms that are identi-
fied, for example, as being related to the De Haerdtl great in-
equality or to a new, potentially interesting inequality. These
788 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions
terms concern an inequality between the three last satellites:
L2 − 3L3 + 2L4 3 . This inequality of a second order of masses
appears in the solution of the semi-major axis and the mean
longitude of Ganymede. The corresponding amplitudes are very
small (respectively, 0.36 and 4.53 kilometers), and the period is
66 days. So, it is not so far from a resonance. Can this inequal-
ity have had an influence upon the evolution of the dynamical
system in the past?
More generally, thanks to the identification performed here,
this synthetic representation holds some analytical information,
in the sense that each term may represent a specific perturbation.
The Galilean system is sometimes referted to with two
more resonances involving the arguments L1 − 2L2 + 1 and
L2 − 2L3 + 2 . The reason is that there are librations of these
arguments. However, this condition is not sufficient. If it were,
such resonances should have brought two new fundamental fre-
quencies (libration frequencies) to the system. Nonetheless, we
succeeded in recognizing all frequencies as a combination of
the fundamental arguments shown in Table 2. Hence, we can-
not confirm the existence of such resonances.
7. Conclusion
We have given the trigonometric series whose arguments are
integer combinations of the fundamental arguments of the
dynamical system, which are also given. According to the
Sampson-Lieske theory, a new representation of the Galilean or-
bital motions deduced from recent ephemerides is now available.
It is a useful base for other dynamical studies on these satellites:
the understanding, in the perturbation sense, of the orbital mo-
tions, and then the history and evolution of the dynamical system
under tidal effects, as well as the rotation of the bodies. However,
the doubtful identification of some terms may still be corrected.
Our work to improve the frequency analysis process, to obtain
better accuracy in the determination of the frequencies, is still in
progress.
3
Not to be confused with the Laplacian resonance, which involves
the three first satellites: L1 − 3L2 + 2L3 .
The present solutions benefited from L1 ephemerides, which
is a former fit to the observations, recently performed in Lainey
et al. (2004b). The tool is now complete and can be easily up-
dated in light of new observations, such as spatial ones (Galileo
spacecraft) or Earth-based ones (mutual events campaigns).
A FORTRAN subroutine computing L1 ephemerides by
means of the present synthetic representation is available on
request.
Acknowledgements. This work benefited from the support of the European
Community’s Improving Human Potential Programme under contract
RTN2-2001-00414, MAGE.
References
Champenois, S. 1999, thesis, Observatoire de Paris
Champenois, S., & Vienne, A. 1999a, Celes. Mech. Dyn. Astr., 74, 111
Champenois, S., & Vienne, A. 1999b, Icarus, 140, 106
Duriez, L. 1982, Celes. Mech. Dyn. Astr., 26, 31
Duriez, L., & Vienne, A. 1997, A&A, 324, 366
Duriez, L. 1990, Le Développement de la Fonction Perturbatrice. In Modern
methods in Celestial Mechanics, ed. D. Benest, & C. Froeschlé (Éditions
Frontières), 35
Duriez, L. 2003, Maple routines for computing disturbing functions,
http://www.univ-lille1.fr/lal/mecanique_celeste.html
Ferraz-Mello, S. 1979, Dynamics of the Galilean satellites, ed. Universidade
de Sao Paulo
Henrard, J. 2005a, Celes. Mech. Dyn. Astr., 91, 131
Henrard, J. 2005b, Icarus, 178, 144
Lainey, V. 2002, thesis, Observatoire de Paris
Lainey, V., Duriez, L., & Vienne, A. 2004a, A&A, 420, 1171
Lainey, V., Arlot, J. E., & Vienne, A. 2004b, A&A, 427, 371
Laskar, J. 1993, Physica D, 67, 257
Lieske, J. H. 1977, A&A, 56, 333
Noyelles, B. 2005, thesis, Observatoire de Paris
Sagnier, J. L. 1975, Celes. Mech. Dyn. Astr., 12, 19
Sampson, R. A. 1921, Mem. R. Astron. Soc., 63
Simon, J.-L., Bretagnon, P., Chapront, J., et al. 1994, A&A, 282, 663
Thuillot, W. 1984, Celes. Mech. Dyn. Astr., 34, 245
Vienne, A., & Duriez, L. 1992, A&A, 257, 331
Vienne, A., & Duriez, L. 1995, A&A, 297, 588
Vu, D. T. 1982, Celes. Mech. Dyn. Astr., 26, 265
V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 1

Online Material
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 2

Table 4. Solution for L1 (mean longitude of Io): L1 = L1 + series expressed in sine, with L1 = 1.4462132960212235 + 3.551552286182 × T .

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
−81.252 282.86691 0.0135848366 ν + 2 0.994D-09 −0.01125
−40.971 247.45358 0.0130341384 ν + 3 −0.155D-09 −0.00026
−37.933 109.32044 0.0030506487 Ψ 0.363D-06 −0.00084
−23.343 85.57787 0.0129389289 ν + 4 0.334D-09 0.00298
−21.253 208.61506 3.5644591050 2L1 − 2L2 0.452D-07 −0.08826
−18.756 104.25814 1.7822295778 L1 − L2 −0.270D-08 0.00526
17.644 150.95338 0.0155711172 ν + 1 0.280D-09 −0.00090
15.715 122.62695 0.0014500977 LS 0.860D-07
−9.894 11.38579 0.0246753155 2ν + 2Ω2 −0.249D-10 −0.00005
−6.766 161.59399 0.0000951962 3 − 4 0.128D-07 0.27849
−5.024 57.02165 7.1289183312 4L1 − 4L2 −0.310D-07 0.03195
−4.601 66.38715 2.6733443704 L1 − L3 −0.783D-08 0.00745
3.681 131.75213 0.0004445681 −Ω2 + Ω3 −0.275D-06 −0.59178
3.470 48.54310 0.0005498008 2 − 4 + Ω3 − Ω4 0.176D-05 −2.25913
3.181 175.58064 0.0006482675 2 − 3 − Ω3 + Ω4 −0.322D-05 10.81254
−2.593 161.29048 8.9111478635 5L1 − 5L2 0.118D-07 0.02653
−2.430 199.16149 8.0200331113 3L1 − 3L3 −0.235D-07 0.02231
−2.245 85.66898 0.0029001291 2LS 0.238D-06
−2.160 265.54919 10.6933774362 4L1 − 4L3 0.142D-07 0.02921
−2.018 282.10444 0.0030554834 2LS − Ω3 − Ω4 0.358D-06 *
1.923 112.21434 0.0229286259 2ν + Ω1 + Ω2 0.800D-06 −0.63457
1.823 90.77307 0.0156771125 2LS + ν − 3 0.285D-05 *
1.590 132.77525 5.3466887181 2L1 − 2L3 0.708D-08 0.01395
−1.325 127.10539 0.0251554892 −LS + 2ν + 2 − Ω4 −0.834D-04 −19.36940 *
1.027 48.88513 0.0025426969 2LS − Ψ + 1 − Ω4 0.147D-05 *

Table 5. Solution for the variable z1 (eccentricity and pericenter of Io). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1751.882 234.33628 −0.0129068641 −ν 0.000D+00 0.00000
264.213 82.86052 3.5515522950 L1 −0.880D-08 0.00140
14.887 121.78955 0.0001272742 3 −0.354D-10 0.00005
8.364 319.91454 0.0000320658 4 −0.652D-09 0.00258
6.179 25.28876 0.0026642534 1 0.000D+00 0.00000
4.168 25.82671 −3.5773660260 −3L1 + 4L2 0.119D-07 −0.01840
−4.086 338.60240 1.7693227079 L2 0.318D-08 −0.00388
−3.506 16.47338 0.8782079195 L3 0.407D-08 −0.00606
2.519 290.72310 7.1160118049 3L1 − 2L2 −0.369D-06 0.66562
−2.204 157.33840 0.0006779610 2 0.124D-07 −0.14646
1.964 281.56981 −5.3595956184 −4L1 + 5L2 0.293D-07 −0.02490
1.939 243.69908 −4.4684808201 −2L1 + 3L3 0.185D-07 −0.02096
−1.592 310.08529 −1.7951364446 −L1 + 2L3 0.558D-08 −0.01257
1.579 177.31174 −7.1418251641 −3L1 + 4L3 −0.983D-10 −0.02822

Table 6. Solution for the variable ζ1 (inclination and node of Io). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
132.609 160.22318 −0.0023150961 Ω1 0.000D+00 0.00000
38.159 60.02914 −0.0005692064 Ω2 −0.234D-10 0.00001
7.145 138.37395 0.0000000000 Ω0 0.000D+00 −0.09677
6.940 191.18949 −0.0001249131 Ω3 0.139D-11 0.00001
2.339 342.17155 −0.0000305612 Ω4 −0.905D-10 0.00880
1.513 48.64339 −0.0252445219 −2ν − Ω2 0.127D-10 0.00002
1.021 318.58614 0.0029003681 2LS −0.646D-09
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 3

Table 7. Solution for the variable a2 (semi-major axis of Europa). The series is expressed in cosine.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
671261.171 0.00000 0.0000000000
64.692 104.25801 1.7822295778 L1 − L2 −0.270D-08 0.00539
23.990 246.38747 2.6733443704 3L2 − 3L3 −0.782D-08 0.00612
−15.246 312.77469 5.3466887181 3L1 − 3L2 0.707D-08 0.01551
13.834 322.12913 0.8911147889 L2 − L3 −0.134D-08 0.00207
−7.834 208.51565 3.5644591656 2L1 − 2L2 −0.155D-07 0.01115
7.652 170.64578 4.4555739318 5L2 − 5L3 0.592D-08 0.01021
−4.666 56.99014 7.1289183312 4L1 − 4L2 −0.310D-07 0.06346
−4.082 161.25599 8.9111478635 5L1 − 5L2 0.118D-07 0.06102
3.474 358.71991 2.7856729212 2L2 − 2L4 0.342D-07 −0.03682
−2.712 247.45298 0.0130341383 ν + 3 −0.319D-10 0.00034
2.617 94.90439 6.2378035398 7L2 − 7L3 −0.271D-07 0.01399
−1.838 265.97440 10.6933772542 6L1 − 6L2 0.196D-06 −0.39399
−1.427 85.58005 0.0129389288 ν + 4 0.425D-09 0.00080
−1.270 9.82442 12.4756070797 7L1 − 7L2 −0.543D-07 0.01939
1.076 282.90788 0.0135848330 ν + 2 0.456D-08 −0.05222
1.034 358.03414 4.1785094280 3L2 − 3L4 0.502D-08 −0.00951

Table 8. Solution for L2 (mean longitude of Europa): L2 = L2 + series expressed in sine, with L2 = 0.3735263437471362 + 1.769322711123 × T .

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
575.703 247.45298 0.0130341383 ν + 3 −0.311D-10 0.00034
305.396 85.58002 0.0129389288 ν + 4 0.426D-09 0.00083
218.089 104.25823 1.7822295778 L1 − L2 −0.270D-08 0.00517
−206.362 282.90958 0.0135848329 ν + 2 0.471D-08 −0.05392
133.070 109.31960 0.0030510121 Ψ 0.000D+00 0.00000
123.114 122.62932 0.0014500979 LS 0.859D-07
−96.285 322.12914 0.8911147889 L2 − L3 −0.134D-08 0.00206
−51.817 246.38746 2.6733443704 3L2 − 3L3 −0.782D-08 0.00613
−42.443 150.95371 0.0155711171 ν + 1 0.417D-09 −0.00123
29.990 312.77450 5.3466887181 3L1 − 3L2 0.707D-08 0.01570
29.306 208.51601 3.5644591656 2L1 − 2L2 −0.155D-07 0.01079
23.439 162.08900 0.0000298857 −Ω4 + Ω0 0.676D-06 −5.99217
−21.864 307.80100 0.0001249523 −Ω3 + Ω0 −0.393D-07 −0.71332
13.816 87.42456 0.0029001316 2LS 0.236D-06
−12.936 170.64581 4.4555739318 5L2 − 5L3 0.592D-08 0.01018
11.279 142.54854 0.0251196097 2ν + Ω2 + Ω3 −0.908D-09 −0.00245
−9.507 167.20958 0.0006493040 2LS + 24 + Ω1 0.976D-07
9.446 11.38600 0.0246753155 2ν + 2Ω2 −0.278D-10 −0.00026
8.857 57.05786 7.1289183312 4L1 − 4L2 −0.310D-07 −0.00426
7.156 305.71256 0.0030233219 2LS − 3 − 2Ω3 −0.402D-06
−6.982 358.71911 2.7856729212 2L2 − 2L4 0.342D-07 −0.03602
6.763 253.75671 0.0005529787 2LS − 4 + Ω1 0.228D-06
6.539 156.48888 8.9111510231 5L1 − 5L2 −0.315D-05 4.82813
−6.354 143.29880 0.0000934783 −Ω3 + Ω4 0.873D-06 7.69205
6.116 90.32784 0.0252444417 2ν + Ω2 + Ω0 0.802D-07 −0.69407
−5.888 88.10350 0.0156763933 2LS + ν − 3 0.356D-05
−5.265 333.05162 0.0259730692 2ν + 3 + 4 −0.172D-08 −0.01745
−5.072 175.24631 0.0004325287 −Ψ + 1 + 3 − Ω2 − Ω3 0.211D-05 −28.70620 *
−4.469 112.24971 0.0229285674 2ν + Ω1 + Ω2 0.858D-06 −0.66994
−4.421 106.67141 0.0260930584 2ν + 23 −0.248D-04 28.23524 *
−3.902 94.90412 6.2378035398 7L2 − 7L3 −0.271D-07 0.01426
3.740 226.69474 0.0025481216 Ψ − 23 + 2Ω3 −0.148D-05 21.42465 *
−3.235 359.36045 1.3928364606 L2 − L4 0.171D-07 −0.01891
2.868 264.82632 10.6933779822 6L1 − 6L2 −0.532D-06 0.75409
2.831 77.40284 0.0030164436 LS + Ψ − 22 − 3 0.153D-05 *
2.531 292.40619 0.0252196582 2ν + Ω2 + Ω4 −0.570D-05 1.13075
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 4

Table 9. Solution for the variable z2 (eccentricity and pericenter of Europa). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
−6282.273 234.33628 −0.0129068641 −ν 0.000D+00 0.00000
200.640 338.60239 1.7693227079 L2 0.318D-08 −0.00387
143.585 121.78957 0.0001272742 3 −0.538D-10 0.00003
132.974 157.19194 0.0006779734 2 0.000D+00 0.00000
81.249 319.93931 0.0000320566 4 0.848D-08 −0.02219
56.187 92.21491 −0.9040216581 −2L2 + 3L3 0.661D-08 −0.00999
55.280 82.86051 3.5515522950 L1 −0.880D-08 0.00141
−21.206 16.47324 0.8782079195 L3 0.407D-08 −0.00592
−19.769 25.82779 −3.5773660260 −3L1 + 4L2 0.119D-07 −0.01948
−18.725 130.08648 −1.7951364497 −3L2 + 4L3 0.107D-07 −0.01275
9.731 167.95660 −2.6862512422 −4L2 + 5L3 0.157D-07 −0.01407
9.334 346.88341 −0.0259410024 −2ν − 3 −0.191D-10 −0.00046
7.275 339.88292 −1.0163502160 −L2 + 2L4 −0.282D-07 0.03251
−5.516 281.57856 −5.3595956184 −4L1 + 5L2 0.293D-07 −0.03365
4.962 148.75621 −0.0258457929 −2ν − 4 −0.465D-09 −0.00078
4.203 356.68305 −7.1418248394 −9L2 + 10L3 −0.325D-06 0.60349
−3.488 311.42179 −0.0264916967 −2ν − 2 −0.500D-08 0.05883
2.920 113.06959 0.3765422106 L4 − 2Ω4 + 2Ω0 0.514D-05 −1.61834
2.825 178.77781 4.4427230757 4L2 − 3L3 − 2Ω4 + 2Ω0 0.512D-05 −1.59203
2.772 243.69827 −4.4684808201 −6L2 + 7L3 0.184D-07 −0.01814
−2.483 300.73159 2.6604375002 2L2 − L3 −0.155D-08 −0.00188
−1.836 72.96507 −8.9240546532 −6L1 + 7L2 −0.860D-07 0.05304
−1.803 43.31370 0.0158069532 2LS + ν − 3 − Ω3 −0.208D-05 *
1.549 111.37726 7.1160114825 7L2 − 6L3 −0.462D-07 0.00844
1.353 335.08976 −2.4091843401 −2L2 + 3L4 −0.238D-05 5.48412
−1.242 153.77170 0.0006823661 LS − Ψ − 4 − Ω1 −0.163D-06 *
1.219 149.24800 6.2248966819 6L2 − 5L3 −0.331D-07 0.00651
−1.201 328.79491 −10.7062843289 −7L1 + 8L2 0.146D-07 −0.04020
1.109 187.11938 5.3337818461 5L2 − 4L3 0.152D-07 0.00393
−1.051 354.02690 −0.0114535888 LS − ν + 3 + Ω3 − Ω0 −0.730D-06 *

Table 10. Solution for the variable ζ2 (inclination and node of Europa). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
2712.225 60.02915 −0.0005692064 Ω2 0.000D+00 0.00000
147.706 191.18947 −0.0001249131 Ω3 0.249D-11 0.00003
111.600 138.28257 0.0000000000 Ω0 0.000D+00 −0.00539
39.623 342.17000 −0.0000305616 Ω4 0.347D-09 0.01035
−7.050 160.22752 −0.0023150966 Ω1 0.514D-09 −0.00434
−6.910 48.64343 −0.0252445219 −2ν − Ω2 0.137D-10 −0.00002
4.873 318.58591 0.0029003677 2LS −0.173D-09
1.235 16.31813 −0.0014500579 −LS −0.126D-06
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 5

Table 11. Solution for the variable a3 (semi-major axis of Ganymede). The series is expressed in cosine.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1070621.016 0.00000 0.0000000000
208.394 66.38716 2.6733443704 L1 − L3 −0.783D-08 0.00744
96.488 322.12917 0.8911147889 L2 − L3 −0.134D-08 0.00203
34.378 74.46116 1.0034433457 2L3 − 2L4 0.346D-07 −0.04047
−18.272 284.25832 1.7822295778 2L2 − 2L3 −0.269D-08 0.00408
16.393 111.65062 1.5051650462 3L3 − 3L4 0.243D-07 −0.01958
10.493 37.22994 0.5017216696 L3 − L4 0.205D-07 −0.01959
8.196 148.92348 2.0068866946 4L3 − 4L4 0.661D-07 −0.08209
−5.904 132.77458 5.3466887181 6L2 − 6L3 0.710D-08 0.01261
−5.434 208.51658 3.5644591656 4L2 − 4L3 −0.155D-07 0.00821
4.353 115.29854 1.7535157350 2L3 − 2LS −0.255D-06 0.43003
4.207 186.15405 2.5086083722 5L3 − 5L4 0.786D-07 −0.10232
−3.111 170.64579 4.4555739318 5L2 − 5L3 0.592D-08 0.01020
2.198 223.38618 3.0103300418 6L3 − 6L4 0.991D-07 −0.12410
−1.794 94.90455 6.2378035398 7L2 − 7L3 −0.271D-07 0.01383
1.677 247.45283 0.0130341383 ν + 3 −0.839D-11 0.00049
−1.639 282.88546 0.0135848349 ν + 2 0.264D-08 −0.02980
1.494 55.40543 0.6269963378 2L3 − 3L4 + 4 −0.713D-05 −0.32458
1.162 260.61772 3.5120517183 7L3 − 7L4 0.113D-06 −0.14529
1.109 9.24780 0.1253136166 L3 − 2L4 + 4 −0.461D-04 8.62270 *
...
0.364 112.32745 0.1123285451 −L2 + 3L3 − 2L4 0.476D-07 −0.03795

Table 12. Solution for L3 (mean longitude of Ganymede): L3 = L3 + series expressed in sine, with L3 = 0.2874089391143348+0.878207923589×
T.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
247.399 122.63008 0.0014500978 LS 0.859D-07
−195.778 247.45287 0.0130341383 ν + 3 −0.568D-11 0.00045
161.918 282.88704 0.0135848348 ν + 2 0.276D-08 −0.03138
−124.590 246.38709 2.6733443704 3L2 − 3L3 −0.782D-08 0.00650
−102.295 85.58049 0.0129389288 ν + 4 0.447D-09 0.00036
87.282 322.12912 0.8911147889 L2 − L3 −0.134D-08 0.00208
−85.780 74.46115 1.0034433457 2L3 − 2L4 0.346D-07 −0.04046
−64.989 37.23009 0.5017216704 L3 − L4 0.197D-07 −0.01974
58.233 160.01306 0.0000298809 −3 + 4 − Ω3 + Ω0 −0.177D-06 −14.79786
−52.381 307.74553 0.0001249528 −Ω3 + Ω0 −0.397D-07 −0.65785
−45.777 104.25834 1.7822295778 L1 − L2 −0.270D-08 0.00506
−32.907 111.71745 1.5051650065 3L3 − 3L4 0.640D-07 −0.08641
−18.176 109.31254 0.0030507678 Ψ 0.244D-06 0.00706
16.565 326.08583 0.0006516407 −ρ − 44 −0.391D-07 −0.35748
−15.553 10.80820 0.1253082718 L3 − 2L4 + 4 −0.407D-04 7.06230 *
−14.523 160.02848 0.0005566368 −ρ − 3 − 34 −0.244D-06 3.82739
−14.416 148.92304 2.0068866946 4L3 − 4L4 0.661D-07 −0.08165
10.227 132.77449 5.3466887181 6L2 − 6L3 0.710D-08 0.01270
9.417 208.51667 3.5644591656 4L2 − 4L3 −0.155D-07 0.00812
8.079 306.87973 0.0000924270 −2LS + 2 − Ω1 − Ω0 0.275D-06 *
−7.617 115.28233 1.7535157644 2L3 − 2LS −0.285D-06 0.44624
6.868 88.95953 0.0029001310 2LS 0.237D-06
−6.769 186.15469 2.5086083722 5L3 − 5L4 0.786D-07 −0.10296
−6.056 142.44956 0.0004483462 −LS − Ψ + 1 − Ω1 + Ω4 −0.754D-06 *
5.628 11.38555 0.0246753155 2ν + 2Ω2 −0.191D-10 0.00019
5.034 170.64584 4.4555739318 5L2 − 5L3 0.592D-08 0.01015
−5.032 55.35749 0.6269971274 2L3 − 3L4 + 4 −0.792D-05 −0.27664
−4.985 206.98165 0.0004363323 LS + 2Ω2 − Ω3 + Ω0 0.352D-06 *
−4.534 112.32691 0.1123285428 −L2 + 3L3 − 2L4 0.500D-07 −0.03741
−4.149 129.60090 0.0251466639 −LS + 2ν − ρ −0.326D-05 *
−3.488 325.79427 0.0017074577 LS − ρ − 2 − Ω3 − Ω4 0.890D-07 *
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 6

Table 13. Solution for the variable z3 (eccentricity and pericenter of Ganymede). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1529.897 121.78960 0.0001272741 3 0.000D+00 0.00000
825.549 319.91861 0.0000320643 4 0.758D-09 −0.00149
634.441 234.33628 −0.0129068641 −ν 0.000D+00 0.00000
219.006 302.02716 −0.1252354408 −L3 + 2L4 −0.160D-07 0.01947
191.119 16.46862 0.8782079244 L3 −0.836D-09 −0.00130
121.194 82.86189 3.5515522950 L1 −0.880D-08 0.00003
−70.530 130.07531 −1.7951364395 −3L2 + 4L3 0.441D-09 −0.00158
53.216 338.59968 1.7693227129 L2 −0.181D-08 −0.00116
−33.873 92.20641 −0.9040216503 −2L2 + 3L3 −0.119D-08 −0.00149
30.813 264.80574 −0.6269571234 −2L3 + 3L4 −0.235D-07 0.03054
−19.458 339.25199 0.3764862399 L4 −0.648D-08 0.00498
11.301 227.57993 −1.1286788041 −3L3 + 4L4 −0.329D-07 0.04600
−7.581 346.88325 −0.0259410024 −2ν − 3 −0.840D-11 −0.00030
−7.580 160.30462 0.0006777426 2 0.231D-06 −3.11268
−6.536 81.08319 −0.8753076942 −L3 + 1 + 3 − Ω3 + Ω0 0.162D-04 −3.38447 *
−6.168 243.68203 −4.4684807883 −6L2 + 7L3 −0.133D-07 −0.00190
−6.136 167.94426 −2.6862512193 −4L2 + 5L3 −0.728D-08 −0.00173
5.171 155.49097 0.0027731330 2LS − 3 −0.396D-07 −0.07450
4.990 190.35365 −1.6304004832 −4L3 + 5L4 −0.440D-07 0.06193
−4.084 148.75671 −0.0258457930 −2ν − 4 −0.437D-09 −0.00128
3.745 90.90874 1.8816512921 3L3 − 2L4 0.118D-07 −0.02073
−3.222 205.81317 −3.5773660056 −5L2 + 6L3 −0.845D-08 −0.00184
−2.648 306.31315 0.2512257684 −L3 + 3L4 − 3 − Ω3 + Ω4 −0.791D-05 4.19169
2.614 269.57365 −0.2504961383 −2L3 + 4L4 − 3 − Ω3 + Ω4 −0.770D-05 3.72084
2.588 197.71646 0.0000000000 3 + Ω3 − Ω0 −0.236D-05 23.01454 *
2.478 317.33703 0.0028683339 LS + 22 − 2Ω4 −0.108D-05 *
2.425 318.61041 0.0014501893 LS −0.555D-08 −0.00737
2.382 153.12540 −2.1321221655 −5L3 + 6L4 −0.519D-07 0.07984
2.243 128.05821 2.3833730193 4L3 − 3L4 −0.252D-07 0.04015
−1.503 53.69648 1.3799296042 2L3 − L4 0.956D-08 −0.01881
1.278 165.36674 2.8850946416 5L3 − 4L4 0.426D-07 −0.05803
−1.197 281.55100 −5.3595955727 −4L1 + 5L2 −0.164D-07 −0.00609
1.186 115.89526 −2.6338438454 −6L3 + 7L4 −0.621D-07 0.09963
−1.110 23.33241 −0.8738575909 −L3 + 1 + 2 − 2Ω2 + Ω3 + Ω0 0.539D-05 −7.91065 *

Table 14. Solution for the variable ζ3 (inclination and node of Ganymede). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1705.791 191.18950 −0.0001249131 Ω3 0.000D+00 0.00000
913.571 138.27696 0.0000000000 Ω0 0.000D+00 0.00022
376.146 342.17492 −0.0000305610 Ω4 −0.238D-09 0.00543
−154.376 60.02914 −0.0005692063 Ω2 −0.842D-10 0.00001
16.841 318.58769 0.0029003665 2LS 0.997D-09
2.694 16.31649 −0.0014500554 −LS −0.128D-06
2.188 101.14211 0.0014501384 LS 0.454D-07
1.921 261.09101 0.0043504622 3LS 0.891D-07
1.458 48.64347 −0.0252445219 −2ν − Ω2 0.127D-10 −0.00006
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 7

Table 15. Solution for the variable a4 (semi-major axis of Callisto). The series is expressed in cosine.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1883133.534 0.00000 0.0000000000
537.835 37.22265 0.5017216817 L3 − L4 0.851D-08 −0.01230
412.594 103.61504 3.1750660413 L1 − L4 0.114D-07 −0.01009
192.606 359.35287 1.3928364698 L2 − L4 0.785D-08 −0.01133
−62.438 74.43094 1.0034433697 2L3 − 2L4 0.106D-07 −0.01025
41.749 40.92573 0.7500722587 2L4 − 2LS −0.159D-06 0.38214
−29.893 111.66879 1.5051650462 3L3 − 3L4 0.243D-07 −0.03775
−14.977 148.89551 2.0068867266 4L3 − 4L4 0.340D-07 −0.05412
−7.689 186.12419 2.5086084022 5L3 − 5L4 0.485D-07 −0.07246
−7.116 278.66101 0.7486221659 2L4 − 3LS −0.250D-06
5.210 8.64176 0.3764591707 L4 − 4 −0.500D-05 10.69809
4.246 296.06433 0.1253067807 L3 − 2L4 + 4 − Ω4 + Ω0 0.870D-05 32.09690 *
−3.938 191.93984 3.0103491233 6L3 − 6L4 −0.190D-04 31.32224 *
3.577 249.65788 0.6269823880 −L3 + 4L4 − LS −0.696D-03 *
3.291 86.37571 2.7986109087 L1 − 2L4 + 4 0.976D-06 −2.11061
−2.560 358.71376 2.7856729335 2L2 − 2L4 0.219D-07 −0.03067
−2.124 260.58738 3.5120517575 7L3 − 7L4 0.736D-07 −0.11495
−1.795 106.47783 1.1287042579 3L4 − 22 + 4 − Ω2 − Ω0 −0.233D-06 −1.47999
1.622 314.39993 6.7395238593 3L1 − 2L2 − L4 0.134D-05 −2.26818
1.619 342.47195 1.0163811590 L2 − 2L4 + 3 − 24 + Ω4 + Ω0 0.167D-05 0.02550
...
0.037 325.97902 0.0006516502 −ρ − 44 −0.486D-07 −0.25067
−0.028 163.85638 0.0005563954 −ρ − 3 − 34 −0.285D-08 −0.00051

Table 16. Solution for L4 (mean longitude of Callisto): L4 = L4 + series expressed in sine, with L4 = 0.3620341291375704+ 0.376486233434× T .

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
1051.926 122.63707 0.0014500979 LS 0.858D-07
−716.686 156.75463 0.0000297297 −3 + 4 − Ω3 + Ω0 −0.256D-07 −11.53943
415.260 37.23060 0.5017216724 L3 − L4 0.177D-07 −0.02025
353.633 103.61820 3.1750660413 L1 − L4 0.114D-07 −0.01325
144.421 359.35978 1.3928364637 L2 − L4 0.140D-07 −0.01824
140.681 74.46111 1.0034433457 2L3 − 2L4 0.346D-07 −0.04042
−73.126 40.84572 0.7500723697 2L4 − 2LS −0.270D-06 0.46215
63.092 307.75076 0.0001249401 −Ω3 + Ω0 −0.270D-07 −0.66308
55.182 111.69205 1.5051650209 3L3 − 3L4 0.496D-07 −0.06101
35.015 83.82945 0.0029001339 2LS 0.234D-06
−32.096 325.97507 0.0006516504 −ρ − 44 −0.488D-07 −0.24672
28.509 164.72259 0.0005564607 −ρ − 3 − 34 −0.681D-07 −0.86672
−28.026 190.91569 0.1253079008 L3 − 2L4 + 4 − Ω4 + Ω0 −0.982D-05 −16.94836 *
24.467 148.92304 2.0068866946 4L3 − 4L4 0.661D-07 −0.08165
−21.866 321.95798 0.0000931663 −LS − 2 − Ω1 + Ω3 − Ω4 −0.579D-06 *
12.469 278.25671 0.7486228617 2L4 − 3LS −0.946D-06
12.313 203.85907 0.0016550514 3LS − 1 − 4 −0.819D-06
11.596 186.15420 2.5086083722 5L3 − 5L4 0.786D-07 −0.10247
8.813 55.35470 0.6269971662 2L3 − 3L4 + 4 −0.795D-05 −0.27385
 V. Lainey et al.: Synthetic representation of the Galilean satellites’ orbital motions, Online Material p 8

Table 17. Solution for the variable z4 (eccentricity and pericenter of Callisto). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
13889.204 319.91712 0.0000320651 4 0.000D+00 0.00000
389.041 339.24735 0.3764862419 L4 −0.851D-08 0.00962
299.394 16.46910 0.8782079244 L3 −0.836D-09 −0.00178
−293.982 121.79592 0.0001272744 3 −0.283D-09 −0.00632
279.842 82.86193 3.5515522950 L1 −0.880D-08 −0.00001
119.593 338.59974 1.7693227129 L2 −0.181D-08 −0.00122
112.866 235.63186 −2.7985797955 −L1 + 2L4 −0.239D-07 0.02017
101.814 317.36333 0.0028683408 2LS − 4 −0.384D-07 −0.07438
−92.198 264.81049 −0.6269571253 −2L3 + 3L4 −0.216D-07 0.02579
62.837 298.32179 −0.3735860173 −L4 + 2LS 0.151D-06 −0.37269
55.709 339.89402 −1.0163502275 −L2 + 2L4 −0.167D-07 0.02141
55.049 301.99244 −0.1252354245 −L3 + 2L4 −0.322D-07 0.05419
37.208 190.89675 0.0000000000 4 + Ω4 − Ω0 −0.150D-05 27.07647 *
−34.565 227.58138 −1.1286788041 −3L3 + 4L4 −0.329D-07 0.04455
17.026 318.60304 0.0014501837 LS 0.000D+00 0.00000
−15.439 190.35557 −1.6304004832 −4L3 + 5L4 −0.440D-07 0.06001
−11.375 80.04373 0.0043191832 3LS + Ω4 − Ω0 0.807D-06 −0.33145
10.714 240.59046 −0.3721359266 −L4 + 3LS 0.244D-06
−7.614 269.33833 −0.2504960289 −2L3 + 4L4 − 4 −0.695D-05 14.83780
−7.420 153.12774 −2.1321221655 −5L3 + 6L4 −0.519D-07 0.07750
6.949 20.17257 1.1265585019 3L4 − 2LS −0.169D-06 0.39227
−5.377 318.57178 −0.0000315849 4 + Ω3 − 2Ω4 + Ω0 −0.141D-06 6.45133
−5.007 14.87965 0.0014182026 LS − 4 −0.839D-07
−3.712 115.89802 −2.6338438454 −6L3 + 7L4 −0.621D-07 0.09687
3.639 294.60685 0.2512311791 −L3 + 3L4 − 4 −0.125D-04 26.77962 *
−3.516 53.67176 1.3799296163 2L3 − L4 −0.256D-08 0.00591
3.171 348.33561 0.7529452084 2L4 − 3 − Ω3 + Ω4 −0.566D-05 −0.62042
−3.144 90.91586 1.8816512864 3L3 − 2L4 0.175D-07 −0.02785
3.073 262.35471 0.0014822429 LS + 4 0.593D-08
−3.043 17.27883 −0.0014180284 −LS + 4 −0.902D-07
−2.643 340.52894 −2.4091866865 −2L2 + 3L4 −0.354D-07 0.04494
−2.265 155.48788 0.0027731373 2LS − 3 −0.439D-07 −0.07141
−2.214 232.51257 −0.7522178950 −2L4 + 22 − 4 + Ω2 + Ω0 0.104D-06 1.74656
−2.034 128.14122 2.3833729650 4L3 − 3L4 0.290D-07 −0.04286
−1.904 78.66637 −3.1355655166 −7L3 + 8L4 −0.811D-07 0.11817
−1.672 289.53911 −0.5016928158 −L3 + L4 + 4 0.319D-05 −6.83233
1.670 287.32356 0.0006835386 −ρ − 34 0.128D-06 −1.67808
−1.543 356.32749 −0.3750361593 −L4 + LS 0.110D-06
1.336 252.77931 −2.4221246132 −L1 + 3L4 − 4 −0.104D-05 2.21256
1.267 182.83092 −0.3706858488 −L4 + 4LS 0.350D-06
1.186 77.89774 1.1251084136 3L4 − 3LS −0.265D-06
1.154 294.51269 7.1160095483 3L1 − 2L2 0.189D-05 −3.12397
−1.051 195.48021 −1.2539396667 −4L3 + 6L4 − 4 −0.669D-05 14.27522
1.002 202.69555 0.0057685340 4LS − 4 0.136D-06
...
−0.892 124.01796 0.0005884547 −ρ − 3 − 24 0.293D-08 −0.24496

Table 18. Solution for the variable ζ4 (inclination and node of Callisto). The series is expressed in complex exponentials.

Amplitude Phase Frequency Identification Freq. diff. Ph. diff.

(km) (deg) (rad/day) (rad/day) (deg)
7235.560 138.27718 0.0000000000 Ω0 0.000D+00 0.00000
4228.368 342.18035 −0.0000305613 Ω4 0.000D+00 0.00000
−490.458 191.18883 −0.0001249131 Ω3 0.291D-10 0.00067
62.541 318.58824 0.0029003769 2LS −0.939D-08
9.364 16.32255 −0.0014500572 −LS −0.127D-06
−9.306 60.02826 −0.0005692030 Ω2 −0.342D-08 0.00089
8.275 101.32342 0.0014501345 LS 0.493D-07
7.086 261.08358 0.0043504645 3LS 0.867D-07
−5.804 115.13221 0.0029313051 2LS − Ω4 + Ω0 −0.376D-06